Single-cell sequencing techniques, such as single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq), have revolutionized our understanding of cellular diversity and function. Genetic and epigenetic factors influence phenotypic heterogeneity in ways that are just beginning to be understood. In this work, we develop methods for inferring mechanisms of biological heterogeneity in single-cell data, with particular applications to cancer biology. First, we develop a kernel archetype analysis method for overcoming noise and sparsity in single-cell data by aggregating single cells into high-resolution cell states. We show that the proposed approach captures robust and biologically meaningful cell states and enables the inference of epigenetic regulation of phenotypic heterogeneity. In the second part of this thesis, we develop methods for linking genotypic and phenotypic information, first by using aggregated single-cell RNA sequencing and a hidden Markov model to infer copy number variation. We demonstrate that aggregation improves copy number inference over existing approaches. We then integrate DNA sequencing with single-cell RNA sequencing to infer copy number profiles in a rapid autopsy of a patient with metastatic pancreatic cancer. We develop a scalable algorithm for inferring phylogenetic relationships between cells from noisy copy number profiles. We show that our approach more accurately recovers phylogenetic relationships between cells and apply it to understand the relationship between genotype and phenotype in metastatic cancer. Finally, we develop a metric for quantifying the extent to which genotype determines phenotype in lineage tracing data. We show that it more accurately quantifies phenotypic plasticity compared to existing approaches. Altogether, these methods can be used to help uncover the mechanisms underlying phenotypic heterogeneity in biological systems.

Aim 1. Accelerated antiretroviral therapy (ART) initiation, including starting ART on the day of HIV diagnosis, has emerged to be one of the approaches to improve ART uptake by shortening or removing some preparatory steps before ART initiation. By doing so, accelerated ART initiation is thought to remove some structural barriers associated with ART initiation process. However, several concerns still need to be addressed, such as whether the expedited process would lead to adverse treatment outcomes after ART initiation. Searched strategy was developed using both MeSH and free text terms relevant to accelerated ART initiation (same-day, immediate, rapid). Exclusion criteria were studies that did not focus on HIV, did not involve HIV treatment, included individuals with HIV aged lower than 12, and contained non-human subjects. Additionally, we excluded articles that were case-reports, qualitative studies, systematic reviews, commentary, points of view, and conference presentations. Four electronic databases (PubMed, Embase, Web of Science, MEDLINE) were used to identify relevant studies published in English between January 2015 and December 2023. Outcomes were retention, viral suppression, pre-ART screening procedures, preferred baseline antiretroviral regimens, additional baseline medications, and adverse events after ART initiation. Two independent researchers were involved in the study selection process. Of 5,455 studies retrieved, 25 studies were included in the review (Cohen’s kappa: 0.88). Six studies reported findings from randomized controlled trials conducted in Lesotho (n=2), Haiti (n=1), South Africa (n=3), and Kenya (n=1), with one study conducted in both South Africa and Keya; 19 studies were observational cohort study from Ethiopia (n=4), West Africa (n=1), Italy (n=2), the United States (n=3), South Africa (n=3), Kenya (n=1), Rwanda (n=1), Sub-Saharan African region (n=1), the United Kingdom (n=1), Turkey (n=1), and China (n=1). The majority of the studies were conducted in urban areas (n=19). Of the 25 included studies, 19 had same-day ART initiation as the intervention or the exposure (three studies measured the time to ART initiation from the day of care engagement, and 16 studies measured it from the day of HIV diagnosis). There was heterogeneity in the pre-ART screening procedures, from relying on symptomatic screening and history assessment to using non-molecular rapid tests to help identify individuals with increased risk of clinical contraindications. Despite this, individuals with symptoms consistent with WHO stage 4 neurological diseases were not eligible for ART. Efavirenz-based ARV was the most regimen reported. The majority of PWH preferred to start ART within 7 days of HIV diagnosis or care engagement (range: 56.5%-86%). Our review suggested mixed results on retention in care and viral suppression after ART initiation, although many studies indicated potential benefits. Despite this, no study reported an association between clinical adverse events, including deaths, and accelerated ART initiation. Our review suggested that accelerated ART initiation can potentially increase ART uptake while not negatively impacting treatment outcomes in some settings. New tools in HIV treatment, such as safer drug regimens and injectable ART, may help improve PWH’s experience and reduce the burden associated with pill burden and frequent clinic visits. Aim 2. Accelerated antiretroviral therapy (ART) initiation has been proposed to address some structural barriers associated with the ART initiation process and improve ART uptake. Despite this, there has yet to be a consensus on how this approach should be implemented, especially concerning the clinical readiness screening procedures. While emerging literature has reported the clinical safety of accelerated ART, limited data are reported from Thailand. Given the heterogeneity of clinical profiles of people with HIV (PWH) in different regions, past studies may not be generalizable to Thailand. Additionally, as different screening procedures affect the time to ART initiation, we need to learn how these procedures impact treatment outcomes. Data were obtained from PWH from 10 ART facilities in six provinces (Chiang Rai, Chiang Mai, Chonburi, Ubon Ratchathani, Songkhla, and Bangkok) in Thailand between July 2017 and July 2019 and followed up until January 2021. All PWH registered in HIV care were included in the analysis, regardless of baseline clinical status. ART facilities were categorized into three models according to the hospital policy on pre-ART laboratory screening procedures: Model A did not consider any lab results at the initiation, Model B considered only CD4 count, and Model C considered other non-CD4 baseline laboratory results. Log-Poisson regression was used to assess the impact of hospital policies on adverse outcomes (deaths, ART discontinuation, loss to follow-up) at months three, six, 12, 18, and 24 after care engagement. Logistic regression was used to examine the impact of hospital policies on viral non-suppression (VNS, HIV-1 RNA>50 copies/mL) at months six, 12, and 18 after ART initiation. Multilevel mixed model was used to account for potential clustering within each hospital policy. Of 10,926 PWH in the dataset, 9,695 (88.7%) were included in this study. Among these, 68% (6,571/9,695), 13% (1,236/9,695), and 19% (1,888/9,695) were in Models A, B, and C, respectively. Both Models A and B had 2 ART facilities each, while Model C had 6 ART facilities. 54.2% (5,257/9,695) self-reported to be men who have sex with men, and the overall baseline median CD4 (IQR) was 168 (129-404) cells/mm3. Compared to Model A, the average risk ratio (95%CI) of adverse events at months three, six, 12, 18, and 24 for Model B was 1.14(1.08-1.20), 1.40(1.31-1.49), 1.19(1.10-1.27), 1.11(1.02-1.21), and 1.32(1.21-1.44), respectively, while it was 1.21(1.16-1.27), 1.76(1.67-1.85), 1.59(1.50-1.67), 1.81(1.71-1.90), and 1.98(1.88-2.10) for Model C, respectively. Of 9,695 PWH, 6,785 (70%) had a confirmed date of ART initiation; 37% (2,513/6,785), 34% (2,332/6,785), and 13% (851/6,785) PWH had information on viral load status at months six, 12, and 24 after ART initiation, respectively. Among these samples, compared to Model A, the average odds ratio (95%CI) of VNS for Model B at months six, 12, and 18 was 0.79(0.59-1.06), 1.06(0.71-1.55), and 1.47(0.49-3.58), respectively, while it was 1.01(0.77-1.32), 0.68(0.40-1.09), and 0.93(0.31-2.22) for Model C, respectively. ART facilities that considered CD4 or any other non-CD4 baseline laboratory results before starting ART had, on average, a higher likelihood of adverse outcomes after the initial care engagement visit and viral non-suppression after ART initiation than ART facilities that did not consider any baseline laboratory result. Aim 3. Clinical screening and psychosocial readiness assessments prior to antiretroviral therapy (ART) initiation are imperative to ensure clinical safety and ART adherence among people with HIV (PWH). However, multiple preparation steps and long wait times associated with ART initiation can contribute to HIV care disengagement and low ART uptake. To address some of the barriers associated with lengthy assessment process, accelerated ART initiation, an approach to start ART on or near the day of HIV diagnosis, has been proposed. Despite this, concerns with the expedited preparation process remain, especially with the PWH’s readiness to have optimal HIV care adherence. This study examined the impact of time to ART initiation on adverse outcomes after care engagement and viral non-suppression (VNS) after ART initiation among PWH in Thailand. Data were obtained from PWH from 10 ART facilities in 6 provinces (Chiang Rai, Chiang Mai, Chonburi, Ubon Ratchathani, Songkhla, and Bangkok) in Thailand between July 2017 and July 2019 and followed up until January 2021. PWH who tested negative for cryptococcal antigen test at baseline and had a confirmed date of ART initiation were included in the analysis and were categorized into three groups based on the time interval between care engagement (defined as the day that PWH first registered at an ART facility) and ART initiation: (1) same day (ART initiation upon the day of care engagement or same day), (2) 1-7 days, and (3) more than 7 days. Log-Poisson regression was used to assess the impact of time to ART initiation on adverse outcomes (deaths, ART discontinuation, and loss to follow-up) at months three, six, 12, 18, and 14 after care engagement. Logistic regression was used to examine the impact of time to ART initiation on VNS (HIV-1 RNA>50 copies/mL) after ART initiation at months six, 12, and 18 after ART initiation. Age, population, hospital policy on pre-ART screening procedures, and baseline CD4 were adjusted in the final models. Of 10,926 PWH in the dataset, 5,528 (50.6%) had complete information on the date of care engagement, negative results for the cryptococcal antigen test, and the date of ART initiation. Among these, 44.23% (2,445/5,528), 38.69% (2,139/5,528), and 17.08% (944/5,528) started ART on the day of, 1-7 days from, and more than 7 days from HIV care engagement visit, respectively. The median age (IQR) was 29 (24-36) and 61% (3,387/5,528) identified themselves as men who have sex with men. The baseline median CD4 (IQR) was 283 (162-412) cells/mm3. Compared to PWH who started ART on the day of HIV care engagement visit, the average risk ratio (RR) of adverse outcomes for those who started ART between 1-7 days at months three, six, 12, 18, and 24 was 0.73(0.60-0.89), 0.66(0.55-0.79), 0.74(0.63-0.86), 0.83(0.71-0.98), and 0.84(0.70-1.01), respectively, while it was 2.27(1.91-2.71), 2.16(1.85-2.52), 1.70(1.46-1.98), 1.93(1.65-2.25), and 2.83(2.44-3.30) for those who started ART more than 7 days, respectively. In the adjusted models, the associations from both groups became statistically non-significant, except for the more than 7 days at month 24 (adjusted RR:1.08; 95%CI:1.04-1.12). Of 5,528 PWH, 29% (1,616/55,28), 36% (1,967/5,528), and 14% (795/5,528) had information on viral load status at months six, 12, and 18 after ART initiation, respectively. Among these individuals, time to ART initiation was determined to have no impact on VNS in both crude and adjusted models. Accelerated ART initiation has the potential to improve ART uptake while maintaining optimal adherence to HIV care. However, HIV programs should recognize and respond to the diversity of needs among PWH to minimize adverse outcomes following ART initiation.

Introduction: Understanding the temporal dynamics of mental health conditions, such as suicidal ideation (SI), is critical for advancing research and clinical interventions. Ecological Momentary Assessment (EMA) is a method of increasing relevance for capturing such data over time, within participants daily lives. However, traditional analytical methods often fail to capture the episodic nature and complex temporal dependencies inherent in EMA mental health data. This project investigates the application of recurrent neural networks (RNNs) to EMA data to improve the prediction and understanding of SI, leveraging their ability to model sequential, high-dimensional data. Methods: Data for this study were drawn from a randomized controlled trial examining the effects of dialectical behavioral therapy (DBT) and SSRI medication on SI in individuals with borderline personality disorder. Participants provided EMA responses multiple times per day over a period at baseline and again post-treatment. RNNs were trained on a portion of each participant’s baseline EMA data with EMA as the outcome, using various baseline and time-varying predictors. Predicted EMA SI values were then generated for a baseline EMA testing dataset, and for the post-treatment EMA period, These predicted SI values were examined to assess the accuracy of the RNN modeling. Baseline testing accuracy was compared to traditional mixed-effects models (MEMs) to demonstrate RNNs feasibility as an alternative for learning and predicting SI time series. Additionally, simulated EMA data was generated in order to describe the data conditions under which RNNs are most useful in modeling EMA data. Furthermore, post-treatment EMA SI predictions were explored to assess the long-term predictive capabilities of RNNs and investigate the prospects of using RNNs to draw causal or mechanistic insights. Results: Key findings underscore the potential of RNNs in mental health research. At baseline, RNNs consistently outperformed MEMs in predicting SI, demonstrating their ability to model complex temporal dependencies and account for within- and between-subject variance. The simulated data analysis highlighted conditions under which RNNs excel, including the use of time-varying predictors and the availability of sufficient longitudinal data, offering guidance for future RNN use. The post-treatment analysis revealed that RNNs continued to provide reasonably accurate predictions, showcasing their robustness even when data were temporally distant from the training period, and following treatment interventions. Furthermore, differences in prediction accuracy between the DBT and SSRI treatment groups suggested that these interventions may uniquely influence SI dynamics in ways that RNN predictions may help to illuminate. Variables associated with prediction error differences provided further insight into treatment-specific mechanisms, highlighting the potential for RNNs to uncover nuanced effects not readily captured by traditional methods. Conclusion: This dissertation advances the understanding of SI as a dynamic and context-dependent mental health outcome. By integrating EMA with RNN-based modeling, it addresses critical gaps in the analysis of temporal mental health data, offering novel insights into both the evolution of SI and the effects of therapeutic interventions. These findings underscore the potential of machine learning techniques to enhance EMA's utility, paving the way for future research and clinical applications aimed at improving mental health outcomes.

Precision medicine aims to tailor medical care and treatment plans based on an individual's characteristics. This dissertation develops machine learning methods to extract meaningful features from digital marker signatures and address the challenges of learning individualized treatment rules using clinical trials and observational studies. The first part of this dissertation proposes a joint nonparametric Bayesian approach that extends the hierarchical Dirichlet process autoregressive hidden Markov model with subject-specific transitions. This model allows for simultaneous learning of latent states across multiple subjects and repeated intensive measurements, facilitating symptom monitoring through wearable device technologies as an objective, low-cost, real-time alternative in movement disorders. The second part introduces a novel approach to integrate the intermediate outcomes from multiple domains through a modified restrictive Boltzmann machine (RBM) model, such that clinical or biological measures can be combined into a personalized composite outcome. This model facilitates the use of interim measures in learning individualized treatment rules for early detection of non-responders and early intervention to improve final outcomes in mental disorders. In the third part, we develop a novel framework for effective and generalizable learning of the individualized treatment effect (ITE) to address the multifaceted nature of treatment responses to mental disorders. This model jointly evaluates multi-domain treatment outcomes and can ensure generalizability across a potentially infinite class of diverse yet clinically relevant outcomes by leveraging a distributionally robust framework and the generalized latent factor models.

A large database of domestic mammal bone measurements from sites across Greenland, Iceland, the Faroe Islands, and the Northern and Western Isles of Scotland is presented. The reasons for variations in bone growth of domestic ungulates are examined in detail; nutrition is identified as a key factor in the determination of adult bone size and shape. Possible sources of variation in bone size in both time and space in the North Atlantic region are identified. Four hypotheses are proposed; firstly that bone dimensions, particularly breadth, will decrease with increasing latitude in the study region; secondly that higher status sites will raise larger livestock than lower status sites within the same time period and region; thirdly the size of domestic mammals in the Northern and Western Isles of Scotland will increase in the Later Iron Age, possibly in relation to increased fodder supply; finally at times of environmental degradation (climatic and/or landscape) domestic mammal size will decrease. The latitude hypothesis could only be partly upheld; there is no evidence for increased size with site status; a small increase in size is noted at some Scottish Iron Age sites and varying results are found for the environmental degradation hypothesis. The results are discussed with particular reference to how changes in the skeletal proportions of domestic mammals affect their human carers and beneficiaries. The potential of further expanding the dataset and integrating biometrical data with other forms of evidence to create a powerful tool for the examination of economic and environmental changes at archaeological sites is discussed.

Combining effect sizes from individual studies using random-effects models are commonly applied in high-dimensional gene expression data. However, unknown study heterogeneity can arise from inconsistency of sample qualities and experimental conditions. High heterogeneity of effect sizes can reduce statistical power of the models. We proposed two new methods for random effects estimation and measurements for model variation and strength of the study heterogeneity. We then developed a statistical technique to test for significance of random effects and identify heterogeneous genes. We also proposed another meta-analytic approach that incorporates informative weights in the random effects meta-analysis models. We compared the proposed methods with the standard and existing meta-analytic techniques in the classical and Bayesian frameworks. We demonstrate our results through a series of simulations and application in gene expression neurodegenerative diseases.

O presente trabalho explora os aspectos taxonomicos, ecologicos, morfologicos, biometricos e biogeograficos dos Arcellinida Kent, 1880 coletados no Parque Ecologico do Rio Tiete, Sao Paulo Brasil. Foram encontrados organismos pertencentes a cerca de 30 taxons nominais, no entanto, a revisao da literatura, novos dados morfologicos obtidos atraves do Microscopio Eletronico de Varredura e medidas biometricas realizadas com grande numero de individuos permitem afirmar que muitos destes taxons estao se referindo a mesma entidade na natureza. Logo, na presente pesquisa sao descritas, com detalhes de distribuicao geografica, morfologia ultra-estrutural, morfometria e ecologia, especies pertencentes a quatro familias e cinco generos: Difflugia corona Wallich, 1864; Difflugia gramen Penard, 1902; Difflugia lanceolata Penard, 1890; Difflugia claviformis Penard, 1899; Difflugia gigantea Chardez, 1967; Centropyxis aculeata (Ehrenberg, 1838); Netzelia wailesi (Ogden, 1980); Lesquereusia modesta Rhumbler, 1895; Lesquereusia mimetica Penard, 1911; Arcella hemisphaerica Perty, 1852; Arcella gibbosa Penard, 1890; Arcella discoides Ehrenberg, 1871 e Arcella brasiliensis Cunha, 1913. Sao discutidas inovacoes taxonomicas para que a comparacao de dados obtidos usando tecnicas atuais com aqueles reportados na literatura tradicional seja feita da maneira mais explicita possivel, de modo a delimitar melhor o conceito taxonomico de cada especie abordada. / The present survey explores the taxonomic, ecologic, morphologic, biometric and biogeographic aspects of the Arcellinida Kent, 1880 collected at the Ecological Park of the Tiete River, Sao Paolo Brazil. Around 30 nominal taxa were identified, however, a review of the literature and new morphologic data obtained via the Scanning Electron Microscope and biometric measures with a large number of individuals allow the inference that many of these taxa are referring to the same natural entity. Therefore, the present work describes species from four families and five genera, along with details about geographic distribution, ultra-structural morphology, morphometry and ecology: Difflugia corona Wallich, 1864; Difflugia gramen Penard, 1902; Difflugia lanceolata Penard, 1890; Difflugia claviformis Penard, 1899; Difflugia gigantea Chardez, 1967; Centropyxis aculeata (Ehrenberg, 1838); Netzelia wailesi (Ogden, 1980); Lesquereusia modesta Rhumbler, 1895; Lesquereusia mimetica Penard, 1911; Arcella hemisphaerica Perty, 1852; Arcella gibbosa Penard, 1890; Arcella discoides Ehrenberg, 1871 e Arcella brasiliensis Cunha, 1913. Taxonomic innovations are discussed in order to make comparison of recent data with those reported on traditional literature a more explicit practice, allowing a better understanding of each species taxonomic concept.

Introdução: A restrição de crescimento intrauterino (RCIU) é uma intercorrência obstétrica de prevalência relevante e altas taxas de morbimortalidade. A ultrassonografia (US) obstétrica ainda é limitada para diagnosticar comprometimento cerebral na RCIU. Por isso, com o intuito de aumentar a acurácia diagnóstica de lesões no encéfalo e comprometimento da criança acometida, surgiram alguns trabalhos utilizando a ressonância magnética (RM), mas com dificuldades técnicas. Sendo assim, são necessários estudos que avaliem o encéfalo de fetos com RCIU e que identifiquem biomarcadores simples de hipóxia crônica e/ou aguda. Objetivos: comparar parâmetros morfométricos mensurados por RM do crânio e encéfalo de fetos com crescimento normal e de fetos com RCIU. Métodos: trata-se de um estudo de coorte prospectivo que incluiu 13 fetos de gestações únicas, com crescimento adequado e 13 fetos de gestações únicas com RCIU, na relação 1 caso:1 controle, de 26 a 38 semanas de idade gestacional (IG) que foram submetidos à avaliação ultrassonográfica para determinação da biometria, volume de líquido amniótico e Dopplervelocimetria fetal e à RM para avaliação de medidas encefálicas e cranianas. Variáveis relacionadas ao tipo de parto, condições do nascimento e resultados perinatais adversos foram obtidas de prontuários médicos. Para análise estatística foram empregados os testes de Wilcoxon e Chi-quadrado. Resultados: as medidas do diâmetro biparietal (DBP) ósseo e cerebral e do diâmetro occipitofrontal (DOF) ósseo de fetos restritos foram menores que as de controles, assim como os percentis desses diâmetros, da circunferência craniana e do DOF cerebral. Observou-se também que a mediana da relação DBP cerebral/cerebelo da população de fetos restritos tendeu a ser menor que a de controles. Além disso, as medidas do líquor cerebroespinhal (LCE) extracerebral e seus percentis também foram menores nos fetos restritos. Também há diferenças nas relações DOF ósseo/LCE, DOF cerebral/LCE, DBP ósseo/LCE e DBP cerebral/LCE entre os grupos de fetos estudados. Além disso, as medidas das distâncias interoperculares axiais direita e esquerda foram significativamente menores nos fetos restritos. Conclusões: podemos concluir que fetos com RCIU possuem medidas cranianas e encefálicas menores que fetos com crescimento adequado, além de haver redução do LCE extracerebral. Estudos de RM fetal com casuística maior, que permitam análise com regressão logística multivariada e aqueles que avaliem comprometimento neurológico das crianças acometidas são necessários. / Introduction: intrauterine growth restriction (IUGR) is an obstetric intercurrence of relevant prevalence and high morbidity and mortality rates. Obstetrical ultrasonography is still limited to diagnose brain impairment in IUGR. Therefore, in order to increase the diagnostic accuracy of brain lesions and impairment of the affected child, some studies using magnetic resonance imaging (MRI) have emerged, but with technical difficulties. Hence, studies that evaluate the brain of fetuses with IUGR and that identify simple biomarkers of chronic and/or acute hypoxia are needed. Objectives: to compare morphometric parameters measured by MRI of the skull and brain of fetuses with normal growth and fetuses with IUGR. Methods: this was a prospective cohort study that included 13 fetuses with normal growth and 13 fetuses with IUGR from singleton pregnancies, in the ratio 1 case: 1 control, from 26 to 38 weeks of gestational age (GI) who underwent ultrasound evaluation to determine the biometry, amniotic fluid volume and fetal Doppler velocimetry and MRI for evaluation of brain and cranial measurements. Variables related to the type of delivery, birth conditions and adverse perinatal outcomes were obtained from medical records. Wilcoxon and Chi-square tests were used for statistical analysis. Results: the measurements of skull and brain biparietal diameter (BPD) and skull occipitofrontal diameter (OFD) of IUGR fetuses were lower than those of controls, as well as the percentiles of these diameters, head circumference and the brain OFD. It has also been observed that the median of the brain BPD/cerebellar diameter ratio of the IUGR fetuses tended to be lower than that of the controls. In addition, measurements of the extracerebral cerebrospinal fluid (CSF) and their percentiles were also lower in IUGR fetuses. There are also differences in the skull OFD/ CSF, brain OFD/ CSF, skull BPD/ CSF and brain BPD/ CSF and extracerebral CSF ratios between the groups of fetuses studied. In addition, measurements of right and left axial interopercular distances were significantly lower in the IUGR fetuses. Conclusions: we can conclude that IUGR fetuses have smaller cranial and brain measures than fetuses with normal growth, besides having reduction of extracerebral CSF. Fetal MRI studies with larger number of subjects, allowing analysis with multivariate logistic regression and those which assess neurological impairment of affected children are needed.

INTRODUÇÂO: O funcionamento da tireóide fetal se inicia em torno de dez semanas de vida embrionária, e está intimamente relacionado ao funcionamento tireoidiano materno. Em gestantes com doenças tireodianas (principalmente hipertireoidismo), a passagem de anticorpos e medicações maternas pode provocar o mau funcionamento da tireóide fetal, acarretando bócio fetal. Além disso, algumas doenças fetais podem cursar com bócio antenatal. O funcionamento inadequado da tireóide fetal pode ter conseqüências severas (restrição de crescimento intra-uterino, craniosinostose, alterações na produção de líquido intra-âmniótico, insuficiência cardíaca ou até óbito fetal). Além disso, o bócio fetal avançado pode funcionar como obstrução à via de parto, podendo acarretar problemas na evolução do parto. A ultra-sonografia da tireóide fetal vem sendo descrita como um bom método para avaliação de tireóide fetal, porém existem poucas curvas de normalidade da tireóide fetal descritas atualmente, nenhuma em população brasileira. O objetivo desse estudo é construir curvas de normalidade do perímetro, área e diâmetro transverso da tireóide fetal em população brasileira através da utilização da ultra-sonografia bidimensional. MÉTODOS: Foram avaliadas 239 gestantes sem doenças sistêmicas e sem história de doença tireoidiana do pré-natal do Hospital das Clínicas da Universidade de São Paulo. Todas as gestantes realizaram dosagem de TSH durante a gestação para descartar doença tireoidiana. A idade gestacional foi calculada pela data da última menstruação, e confirmada por ultrasonografia de primeiro ou segundo trimestre. Foram construídas curvas de normalidade do perímetro, área e diâmetro transverso da tireóide fetal. Das 239 pacientes inicialmente avaliadas, 43 (18%) foram excluídas. A prevalência de hipotireoidismo subclínico entre as gestantes foi de 0,9%, e a de hipotireoidismo franco de 2,2%. Em 5,4% das pacientes não foi possível a visualização adequada da tireóide fetal. Foram incluídas 196 pacientes. A avaliação foi realizada por dois operadores independentes. Foram realizadas três medidas de cada parâmetro, e considerada a média dos valores para a construção das curvas. Para o cálculo da variação intra-observador, foram avaliadas 159 pacientes e realizadas três medidas subseqüentes de cada parâmetro. Para o cálculo da variação inter-observador foram avaliadas 34 pacientes, nas quais cada operador realizou uma medida de cada parâmetro. RESULTADOS: Foram construídas curvas de normalidade do perímetro (P), área (A) e diâmetro transverso (DT) da tireóide fetal em relação à idade gestacional (IG) em nossa população de gestantes. As equações que melhor representaram a média esperada por idade gestacional foram equações lineares: P = 0,146 x IG; A = -1,289 + 0,085 x IG; DT = 0,054 x IG. / INTRODUCTION: The functioning of fetal thyroid initiates around ten weeks of embryonic life, and is intimately related to maternal thyroid functioning. In pregnant women with thyroid disease (especially hyperthyroidism), maternal antibodies and medications provoke malfunctioning of the fetal thyroid gland, causing fetal goiter. Moreover, primary fetal anomalies may course with antenatal goiter (i.e.: congenital fetal hypothyroidism). The inadequate functioning of fetal thyroid causes severe consequences such as intrauterine growth restriction, craniosinostosis, altered production of intra-amniotic liquid, cardiac insufficiency and even fetal death. Moreover, large fetal goiters may cause a mass effect, causing difficulties during vaginal delivery. Ultrasound evaluation of fetal thyroid has recently been described as a sensible method for fetal screening of thyroid anomalies. Few normality curves have been described until now, none of them in Brazilian fetuses. The aim of this study was to build normality curves of fetal thyroid perimeter, area and transverse diameter through the use of bidimensional ultrasonography. METHODS: 239 pregnant women without systemic disease and without previous thyroid disease were evaluated in the prenatal care unity of the Hospital of the Clinics of the University of São Paulo. All women had TSH measured during pregnancy in order to exclude thyroid disease. Gestational age was calculated by the date of the last menses, and confirmed by first or second trimester ultrasonography. Normality curves of fetal thyroid perimeter, area and transverse diameter were constructed. From the 239 patients initially evaluated, 43 (18%) were excluded. The prevalence of thyroid hormone disorders was 0.9% for subclinical hypothyroidism, and 2.2% for hypothyroidism. One patient had a TSH value below normal. In 5.4% of patients the adequate visualization of fetal thyroid was not possible. The evaluation was carried out in 196 patients by two independent operators. Three different measures of each parameter were performed. The average values were used to build the curves. Intra-observer variation analysis was made using 159 patients in whom three measures were made for each parameter. For the inter-observer variation, 34 patients were evaluated. RESULTS: Normality curves of the perimeter (P), area (A) and transverse diameter (TD) were constructed in relation to gestational age (GA) in our population. The 95% confidence interval was calculated. The equations that better represented the average value expected for each gestational age were linear regressions: P = 0,146 x GA; A= -1,289 + 0,085 x GA; TD = 0,054 x GA. The method was considered to be reproductive.

Change-point detection in hazard rate function is an important research topic in survival analysis. In this dissertation, we firstly review existing methods for single change-point detection in piecewise exponential hazard model. Then we consider the problem of estimating the change point in the presence of right censoring and long-term survivors while using Kaplan-Meier estimator for the susceptible proportion. The maximum likelihood estimators are shown to be consistent. Taking one step further, we propose an counting process based and least squares based change-point detection algorithm. For single change-point case, consistency results are obtained. We then consider the detection of multiple change-points in the presence of long-term survivors via maximum likelihood based and counting process based method. Last but not least, we use a weighted least squares based and counting process based method for detection of multiple change-points with long-term survivors and covariates. For multiple change-points detection, simulation studies show good performances of our estimators under various parameters settings for both methods. All methods are applied to real data analyses. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2014. / FAU Electronic Theses and Dissertations Collection